Next Article in Journal
A Channel Centerline-Based Method for Modeling Turbidity Currents Morphodynamics: Case Study of the Baco–Malaylay Submarine Canyon System
Previous Article in Journal
Modeling of Marine Assembly Logistics for an Offshore Floating Photovoltaic Plant Subject to Weather Dependencies
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JMSE
Volume 13
Issue 8
10.3390/jmse13081494
jmse-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Quantitative Risk Assessment of Liquefied Natural Gas Bunkering Hoses in Maritime Operations: A Case of Shenzhen Port

by
Yimiao Gu
1,
Yanmin Zeng
2 and
Hui Shan Loh
3,*
1
Department of Electronic Business/Institute of Digital Business & Intelligent Logistics, South China University of Technology, Guangzhou 510641, China
2
Shenzhen Port Group, Shenzhen 518081, China
3
Logistics and Supply Chain Management Programme, School of Business, Singapore University of Social Sciences, Singapore 599494, Singapore
*
Author to whom correspondence should be addressed.
J. Mar. Sci. Eng. 2025, 13(8), 1494; https://doi.org/10.3390/jmse13081494 (registering DOI)
Submission received: 13 June 2025 / Revised: 29 July 2025 / Accepted: 30 July 2025 / Published: 2 August 2025
(This article belongs to the Section Ocean Engineering)
Browse Figures
Versions Notes

Abstract

The widespread adoption of liquefied natural gas (LNG) as a marine fuel has driven the development of LNG bunkering operations in global ports. Major international hubs, such as Shenzhen Port, have implemented ship-to-ship (STS) bunkering practices. However, this process entails unique safety risks, particularly hazards associated with vapor cloud dispersion caused by bunkering hose releases. This study employs the Phast software developed by DNV to systematically simulate LNG release scenarios during STS operations, integrating real-world meteorological data and storage conditions. The dynamic effects of transfer flow rates, release heights, and release directions on vapor cloud dispersion are quantitatively analyzed under daytime and nighttime conditions. The results demonstrate that transfer flow rate significantly regulates dispersion range, with recommendations to limit the rate below 1500 m3/h and prioritize daytime operations to mitigate risks. Release heights exceeding 10 m significantly amplify dispersion effects, particularly at night (nighttime dispersion area at a height of 20 m is 3.5 times larger than during the daytime). Optimizing release direction effectively suppresses dispersion, with vertically downward releases exhibiting minimal impact. Horizontal releases require avoidance of downwind alignment, and daytime operations are prioritized to reduce lateral dispersion risks.

1. Research Background

As a key area of international energy consumption and one of the main sources of carbon emissions, the transportation industry exhibits significant differences in the environmental efficiency of different transportation modes. According to data from the World Bank, shipping accounts for approximately 3% of global greenhouse gas emissions and 15% of other air pollutants annually [1]. Against this backdrop, liquefied natural gas (LNG), as a clean energy source that is conducive to carbon reduction, has attracted considerable attention due to its large-scale application in the energy sector. In 2021, the Clydebank Declaration on Green Shipping Corridors officially proposed the concept of “green shipping corridors” at the international level, namely “zero emission routes”. Among them, nine corridors in the container and ferry sub-markets are considering using natural gas as fuel [2]. According to data from the General Administration of Customs of the People’s Republic of China [3], in 2023, China’s LNG import volume reached 71.32 million tons (approximately 9.84 billion cubic meters), making China the world’s largest importer of LNG. With the growth in LNG imports, China has also accelerated the construction of clean energy infrastructure. China’s first offshore international LNG bunkering center was officially established in 2020 at Yantian Port in Shenzhen, Guangdong Province [4].
However, LNG presents certain hazards, such as flammability, explosiveness, low temperature, and asphyxiation risks [5]. One of the most common accidents is the ignition and explosion resulting from the mixture of natural gas and air when exposed to an ignition source. LNG has a rapid flame propagation speed, and its combustion rate is significantly high, approximately twice that of gasoline. The flame temperature is high and the radiative heat is intense, leading to the potential for large-scale fires. Moreover, LNG fires are prone to re-ignition and re-explosion, making them difficult to extinguish. After release, liquefied natural gas forms a white vapor cloud due to the heat transfer between the ground and air. Natural gas is lighter than air, so it spreads rapidly in the air. After encountering a fire source, the flame spreads to areas where oxygen can reach. The burning vapor prevents the further formation of vapor clouds and then forms stable combustion. When the mixing ratio of natural gas and air is within the range of 5 vol% to 15 vol%, an explosion occurs. Therefore, it is necessary to study and explore the scenarios and consequences of LNG leakage risks. On 2 November 2020, a fire occurred during the construction of the Phase II project involving simultaneous loading of lean and rich LNG at the Tieshan Port (Linhai) Industrial Area in Beihai City, Guangxi Zhuang Autonomous Region [6]. The direct cause of the accident was the opening of an isolation valve, which caused LNG to spray from a cut pipe. The resulting LNG vapor and air mixture ignited upon encountering a potential ignition source. The accident resulted in 7 fatalities, 2 serious injuries, and direct economic losses amounting to RMB 20.293 million.
In accordance with the requirements of the “Guidelines for Liquefied Natural Gas Fuel Refueling Operations” [7] regarding quantitative risk assessment in refueling operations, simulation quantitative analysis should be carried out on the consequences of leakage accidents during the bunkering process of LNG bunkering ships. The main component of LNG is methane (molecular formula CH4), which is odorless, colorless, non-corrosive, and non-toxic. It is a product of natural gas liquefied at a certain pressure through deep freezing, with a volume of about 1/590 of the volume of natural gas under standard conditions (101.325 kPa, 0 °C). It is suitable for long-distance transportation and large-scale storage. The natural gas generated at room temperature after LNG gasification has a lower density than air and is prone to dispersion after leakage. When conducting risk analysis of the consequences of LNG release, models can be used to assess the risk of the event, and release consequences can also be analyzed through on-site or wind tunnel experiments, theoretical research, and numerical simulation. Since the late 1960s, some research institutions in the United States have conducted on-site experiments on LNG release and dispersion. Famous and representative on-site experiments include the Falcon experiment [8], Coyote experiment [9], Maplin Sands experiment [10,11], Burro experiment [12], Avocet series experiment, Capistran experiment, and Gadila Jettision experiment [13]. The measured data obtained from the above experiments provide strong support for the subsequent study of the consequences of LNG leaks, fires, and other accidents, as well as for the construction of mathematical models, such as the Burro series experiments commonly used by scholars for model validation [14,15]. However, due to the large amount of manpower and funds required for field experiments, few scholars have carried out large-scale field experiments in recent years, in addition to studying the application of chemical materials in LNG accident scenarios, such as the inhibitory effect of high expansion foam on the concentration of LNG leakage gas [16]. More scholars and research institutes are conducting theoretical research on the risk analysis and dispersion models of LNG release.
In the late 1970s, supported by a series of theories and experiments, researchers summarized and proposed many empirical computational models. Hu [17] considered five factors, including frequency and severity during navigation and accident characteristics, and proposed a new model based on relative risk assessment. Luketa Hannlin [10] described explosion, pool boiling, and rapid phase transition phenomena, discussed prediction models for dispersion and hazardous distance, and demonstrated that parameters such as LNG release, dispersion and evaporation, and pool fires have a significant impact on hazardous distance. Hissong [18] considered the temporal variation of LNG spillage and evaporation when mixed with water, analyzed these processes using material and energy balance methods, and concluded that the degree of harm caused by LNG release into water is closely related to the size of the liquid pool formed. They also estimated the heat transfer results by accounting for turbulence. Lehr W’s study [19] compared the possible similarities and differences between accidents involving accidental release, dispersion, and fires caused by different hydrocarbon fuels, including LNG, from ships. The study also included quantitative estimation of characteristic parameters such as spreading speed, duration, maximum floating oil size, combustion speed, smoke, and thermal radiation. Sun [20] proposed a new model based on the Gaussian plume model using the virtual source method and fluid settling theory, and discussed the effects of release direction, release aperture, and tunnel cross-section on concentration distribution. The authors of [21] proposed a Dynamic Risk Assessment Model for the Domino Effect in Urban LNG Stations (DRADL), which uses a set of physical processes and effect models to simulate event processes. Nubli [22] considered the structural integrity and damage caused by collisions of the target ship and used CFD methods to discuss the gas diffusion characteristics of LNG refueling ships and the influence of low-temperature flow on the structure of refueling ships. Luan [23] proposed a multiphase CFD model combined with an evaporation model to simulate the complete overtopping process (including LNG overtopping, LNG dispersion, LNG evaporation, and vapor dispersion) and analyzed the effects of embankment structure on wind field, overtopping rate, LNG dispersion, and LNG vapor cloud dispersion. Xie [24] developed a numerical model of a release jet and conducted theoretical analysis on a series of release situations at different heights of tank openings and under varying internal pressures and external temperatures. The study explored the horizontal stretching of the release vapor cloud on the ground and its impact on the high-pressure steam jet near LNG storage tanks. These theoretical methods have laid a solid foundation for numerical simulation and software simulation. Other empirical computational models include Gaussian model [20], FEM3 model [25], Slab model [26], UDM, BM model, etc. At the beginning of the 21st century, with the rapid development of the computer industry, research institutions and scholars in various countries have developed various numerical models specifically for simulating release, dispersion, and combustion. After continuous improvement and development, simulation has become a new hotspot in the field of LNG shipping safety research [27].
Scholars have established a simulation model for LNG release accidents to estimate the leakage, dispersion, and evaporation rates in open water areas, while identifying hazardous areas based on thermal radiation levels [28]. It has also been determined that other materials, such as LPG and gasoline, with the same confidence level as LNG, have the same effects on fuel release, dispersion, vaporization, and pool fires. Consequently, the same basic equations and principles are applicable. Additionally, a comprehensive model was constructed to predict the dynamic leakage of LNG and product oil tankers [29]. By specifying the characteristics of the pool and establishing variables such as pool fire, pool fire area, duration, and heat release rate, it has been determined that the number and speed of leaks depend on the size and location of the rupture, ultimately reaching a static equilibrium state. Compared to other petroleum products, LNG has a larger leak volume, larger pool fire area, higher combustion heat rate, and faster combustion speed. Another scholar [30] used a two-phase, multi-species model to simulate the dispersion process of LNG vapor clouds and comprehensively considered three common water phase change treatment methods to explore the influence of water vapor phase transition on LNG vapor cloud dilution under different relative humidity conditions. These methods include ignoring water phase changes, considering water phase changes using the Lee model, and considering water phase changes using the Sun model to explore the role of water vapor phase transition in LNG vapor cloud dilution. Wesierski [31] used a combination of experimental and numerical simulation methods to study the propagation characteristics of natural gas vapor clouds leaked from LNG transport tankers for 2–4 min. Bozzolasco [32] used simulation software to simulate the risk consequences of release at LNG receiving stations and determined the impact level based on gas cloud concentration and thermal radiation.
The most commonly used software for simulating LNG leaks currently includes professional accident consequence analysis software PHAST 8.7 [33,34], computational fluid dynamics (CFD) software Fluent [35,36,37,38], and FLACS [39,40,41]. FLACS is CFD software specifically designed for gas explosions, fires, and smoke diffusion. Based on turbulent combustion models and gas dynamics equations, it is suitable for simulating explosions and fires after combustible gas leaks. It can be used to compare different operating and environmental conditions, leak hole diameter, leak rate, wind speed, and wind direction factors; study the characteristics of fuel leak dispersion; and determine the range of safe zones [39,40,41]. Fluent is based on CFD and simulates complex fluid flow, heat transfer, and chemical reactions by solving Navier–Stokes equations and other conservation equations, such as energy, mass, and momentum. It can simulate the dispersion characteristics of a vapor cloud under different environmental conditions (such as wind speed, wind angle, temperature, cofferdam height, leakage amount) [35,37,38], assess its combustion and explosion hazards to ships, personnel, and equipment [36], and propose hazard mitigation measures and safety design optimization schemes, such as sensitivity analysis of heat expansion (HEX) foam height and quantitative assessment of ship lift accidents [37].
PHAST uses the Unified Dispersion Model (UDM) to model jets and dense, buoyant, and passive dispersion, including droplet rainout and re-evaporation. It combines simplified equations of fluid dynamics, thermodynamics, and chemical reactions that are suitable for simulating large-scale leaks, dispersion, fires, and explosions. The research conducted using PHAST mainly focuses on simulating LNG release, dispersion, and vapor cloud formation [33,34], analyzing the effects of different environmental conditions (such as wind speed, wind direction, and terrain) on the dispersion range and concentration of LNG vapor clouds [33,42], and evaluating the potential risks of fire and explosion [31,40]. PHAST is currently the most comprehensive software for process industry hazard analysis and quantitative risk assessment [33]. Its computational model has been validated by DNV GL Spadeadam [43], the world’s largest fire and explosion laboratory, and has been recognized by the State Administration of Work Safety in China. PHAST is based on empirical models and simplified physical equations, with fast calculation speed, a user-friendly interface, and high cost-effectiveness. In contrast, Fluent focuses on high-precision CFD simulations, which require high computational resources and are time-consuming; FLACS focuses on explosion and fire simulation and is suitable for complex geometric environments. The speed, comprehensiveness, and ease of use of PHAST make it an ideal tool for LNG hazard consequence analysis, particularly suitable for rapid risk assessment in large-scale scenarios. The comparison of the three hazard analysis tools is shown in Table 1.
According to the guidelines [7] published by the China Classification Society, when conducting quantitative risk analysis on LNG, the scope and degree of disaster impact should be calculated based on vapor cloud dispersion, thermal radiation, and explosion shock waves. Existing studies on LNG release and dispersion mostly consider the effects of environmental factors (such as wind speed and direction, atmospheric stability, etc.), tank properties (volume, pressure, etc.), and tank release aperture. Recent studies have highlighted the significance of safety zones and ventilation strategies in the applications of alternative fuels, e.g., ship-to-ship LNG bunkering [44] and hydrogen-fueled ships [45]. There is relatively little research on hose release during bunkering operations [46]. However, during the process of LNG bunkering, release may occur at different heights and in different directions at the hose or hose connection due to differences in safety devices and the size configuration of the receiving vessel, or there may be different transfer flow rates. The impact of these important operational parameters on the vapor dispersion of LNG release has not been thoroughly studied.
This study aims to provide a reference for safe LNG bunkering operations, analyze the consequences of hose release during LNG bunkering operations, and examine how operational parameters affect the dispersion and impact range of LNG vapor clouds in the event of an accidental release during LNG bunkering operations at Yantian Port in Shenzhen. This study selected the transfer flow rate, release height, and release direction as parameters in the simulation section to investigate the influence of these parameters on the LNG release range. The research focuses on the following three main sub-topics: (i) transfer flow rate, (ii) release height, and (iii) release direction. These factors are examined through PHAST simulation using real-world meteorological data. This study offers a novel contribution by focusing on dynamic operational variables such as transfer flow rate, release height, and direction under both daytime and nighttime conditions, which are not extensively covered in the literature. In addition, this study uses PHAST to simulate realistic STS scenarios using real-world meteorological data and generates quantitative safety thresholds. The findings complement existing studies and enhance the understanding of their application to risk-based operational decision-making.

2. Methods

This section establishes the methodological framework for quantifying the risks associated with LNG bunkering operations, particularly focusing on potential hose failure scenarios. The method systematically integrates multiple critical dimensions to ensure a robust risk assessment. First, typical LNG bunkering operations and container configurations are characterized to define the operational context. Risk parameters are then identified, and potential release scenarios are developed to model accident conditions. A specialized simulation tool is employed to analyze the consequences of these scenarios, incorporating environmental factors such as weather conditions and dispersion surface properties. Finally, key operational parameters, including transfer flow rate, release height, and direction, are specified to complete the technical foundation for the quantitative analysis. Together, these components provide a comprehensive basis for evaluating LNG bunkering hazards under realistic operating conditions.

2.1. LNG Bunkering Scenarios and Container Parameters

This study selects the Yantian International Port Area in Shenzhen as the operation site and chooses the most common ship-to-ship STS bunkering mode as the research scenario. An 8500 m3 LNG bunkering ship is used as the study object, and a simulated release scenario is established based on PHAST 8.7 software version 8.7. Parameters are selected to simulate the dispersion of LNG from the release point in the atmosphere, and the influence of different operational parameters on the degree of spatial dispersion is evaluated. This study analyzes the most common scenario of ship-to-ship (STS) bunkering mode, as shown in Figure 1.
Due to the fact that the selected location for this study is Yantian Port in Shenzhen, China, the parameters mainly refer to the “Guidelines for Liquefied Natural Gas Fuel Refueling Operations” [7] published by the China Classification Society. LNG is generally loaded into storage tanks for transportation at a temperature of −162 °C and 1 standard atmospheric pressure. In order to simplify the bunkering operation and reduce the requirements for gas system, heater, and cooler balance, the fuel storage tank of the bunkering ship and the receiving tank of the receiving ship are kept under the same temperature and pressure conditions. This study assumes that LNG is transferred from the LNG storage tank carrier of an 8500 m3 LNG bunkering ship to the storage tank of the receiving ship, and the subsequent PHAST model is also modeled based on this scenario. The actual composition of LNG is relatively complex. In order to facilitate analysis, methane, the main component of LNG, is used as the material for simplified simulation calculations. This section simulates LNG at a temperature of −162 °C and a pressure of 101 kPa. The conveying process uses a transfer hose with a length of 40 m and a diameter of 20 cm.

2.2. Risk Parameters and Release Scenarios

In the LNG ship-to-ship bunkering scenario, the main storage tank equipment includes the loading and unloading arm and hose. Key bunkering operations include the handling, connection, and disconnection of the loading and unloading arm and hose. Release and rupture of connecting hoses are the main risk factors for LNG bunkering leaks [47]. Meanwhile, The Netherlands Organization of Applied Scientific Research (TNO) Purple Book [48] also indicates that the probability of release caused by hose rupture and loading/unloading arms (6 × 10−4) is higher than the probability of direct release from sealed storage tanks (6 × 10−5). This paper focuses on the transfer hose and selects three operational parameters for study: transfer flow rate, release height, and release direction. In the process of quantitative analysis of hazardous consequences, the study of three operational parameters is based on the scenario of a rupture at the manifold of a 20 cm diameter hose from the receiving ship. At present, the international release aperture range for LNG bunkering hoses is 10% DN (DN is the nominal diameter) to 50 mm [49]. This study selects 10% DN, which is 20 mm, as the parameter for hose release aperture.

2.3. Simulation Tool

Based on years of experience in safety management and engineering technology, Det Norske Veritas (DNV) has developed the Process Hazard Analysis Software Tool (PHAST version 8.7). As quantitative risk assessment software, PHAST is mainly suitable for hazard analysis and calculation of flammable, explosive, toxic, and harmful materials such as petroleum, petrochemical, and natural gas. PHAST is one of the most widely used professional software programs due to its simple operation and intuitive results [50]. The applicability of its simulation method and the accuracy of its results have been verified through multiple experiments, with a high degree of agreement and reliability.
PHAST software simulation mainly conducts quantitative risk analysis on the scope and severity of possible accident consequences by setting up accident scenarios. The main parameter set includes grid parameters, surface parameters, pool vaporization parameters, discharge parameters, dispersion parameters, fire parameters, explosion parameters, and general parameters. In addition, it includes model parameters (calculation background and methods, buildings and terrain, time, etc.), material parameters (types, components, attributes, etc.), and weather parameters (wind speed, temperature, atmospheric stability, etc.). The impact range and severity are ultimately characterized by the effects of four modules: discharge, dispersion, fire and explosion, and toxicity.

2.4. Weather Parameters and Dispersion Surface

The study was conducted in Shenzhen, Guangdong Province, China, which is a coastal city located in the southern part of the province adjacent to Hong Kong. Shenzhen is located south of the Tropic of Cancer, with longitudes ranging from 113°46′ E to 114°37′ E and latitudes ranging from 22°27′ N to 22°52′ N. This study selects typical meteorological data from Shenzhen in 2023, sourced from the Meteorological Bureau of Shenzhen Municipality, as the weather parameters for the simulation scenario [51]. Ship-to-ship bunkering operations are carried out at sea, with a water surface roughness set at 0.2 mm. Table 2 shows the typical meteorological data. The study used different stability classes for day and night to reflect more accurate and realistic operational conditions.

2.5. Operational Parameters

This study changes three operational parameters—transfer flow rate, release height, and release direction—to study their impact on the dispersion range of the released LNG cloud. The set values of the operational parameters are shown in Table 3.

2.5.1. Transfer Flow Rate

In the process of bunkering operations, in order to minimize transportation time, higher flow rates are generally prioritized. But when the hose ruptures and leaks, a higher transfer flow rate may lead to more LNG release. Therefore, it is necessary to determine the balance point between conveying efficiency and release risk. According to the Guide to Manufacturing and Purchasing Hoses for Offshore Moorings [52], the maximum allowable flow rate inside the hose is 21 m/s. For a hose with a diameter of 8 inches (20 cm), the allowable flow range is 0~2200 m3/h. Select four conveying flow rates between 500 and 2000 m3/h at intervals of 500 m3/h for analysis. This step is intended to provide reference data for selecting transfer flow rates that balance efficiency and risk.

2.5.2. Release Height

Due to the different model parameters of various types of receiving ships, the height of the injection hose manifold also varies. The common height of the injection hose manifold is generally 5–20 m. This study analyzes release heights at four vertical distances of 5–20 m from sea level, with intervals of 5 m, to provide a theoretical basis for the installation of pipelines and the placement of transfer hoses on the receiving ship.

2.5.3. Release Direction

When a hose releases, there may be various holes that rupture in different directions. The different fixing methods and positions of hoses can also lead to different release directions. This study selects four directions for analysis: horizontal, 90° up, 90° down, and 45° down. Based on the severity of hazardous consequences in different directions, the induction parameters of the gas detection device can be optimized and set.

3. Results and Discussion of PHAST Simulation

This section presents the comprehensive simulation results obtained through PHAST modeling, systematically examining LNG dispersion characteristics based on three critical operational parameters. The analysis focuses on both horizontal dispersion patterns (depicted through 25,000 ppm vapor cloud footprints) and vertical dispersion profiles for each scenario. The study reveals how the transfer flow rate, release height, and release direction distinctly influence vapor cloud behavior. The dual-perspective approach not only quantifies hazard zones but also provides three-dimensional insight into dispersion mechanics, enabling practical safety recommendations for bunkering operations.

3.1. Transfer Flow Rate

When the transfer flow rates are 500 m3/h, 1000 m3/h, 1500 m3/h, and 2000 m3/h, and the concentration reaches 25,000 ppm, the dispersion distance and width are as shown in Figure 2. As shown in Figure 2, with the increase in the transfer flow rate, the distance and width of vapor cloud dispersion significantly increase, but the variation pattern of the vapor cloud also varies under different weather conditions. Table 4 shows the dispersion distance and footprint for different transfer flow rates.
At night, methane forms clouds in the upwind direction. With the increase in the transfer flow rate, the width of the vapor cloud in the upwind direction gradually decreases. The low wind speed (2 m/s) at night and stable atmospheric conditions (Class E stability) significantly inhibit turbulent mixing, which limits the dispersion ability of vapor clouds in the upwind direction. With the increase in the transfer flow rate, the initial momentum and volume of the vapor cloud increase, but its dispersion energy is more distributed in the downwind direction through lateral expansion, while the dispersion in the upwind direction is directly blocked by the incoming wind, resulting in the gradual reduction in the width of the vapor cloud in the upwind direction.
In the downwind direction, when the transfer flow rate is 500 m3/h, the dispersion width of the vapor cloud in the daytime and at night shows little difference. However, when the transfer flow rate increases to 1000 m3/h, the downstream dispersion distance of the vapor cloud in the daytime increases significantly, and the cloud footprint increases from 33,870 m2 to 128,903 m2, an increase of about 2.8 times. When the transfer flow rate is lower than 1500 m3/h, the dispersion distance of the vapor cloud in the daytime is smaller than that at night, but when the transfer flow rate is 2000 m3/h, the dispersion distance of the vapor cloud in the daytime is larger than that at night.
At night, when the transfer flow rate increases from 500 m3/h to 1000 m3/h, the maximum downwind distance also increases from 530.36 m to 1610.13 m, and the cloud footprint increases from 54,719 m2 to 458,254 m2, with an increase of about 7.4 times. However, when the transfer flow rate increases to 1500 m3/h and 2000 m3/h, the maximum downwind distance decreases to 1493.49 m and 1416.16 m; the cloud footprint is 654,081 m2 and 666,580 m2. With the flow rate increasing from 1500 to 2000 m3/h, the footprint increases by only about 0.02 times. This is because when the transfer flow rate is large, the initial momentum of the steam is high, but because the downwind dispersion at night is limited, the dispersion energy of the steam spreads more in the horizontal direction, rather than continuing downwind.
In the daytime, when the transfer flow rate increases, the dispersion distance and range of the vapor cloud increase significantly. At night, with the increase in the transfer flow rate, the dispersion range of the initial vapor cloud also increases significantly. When the transfer flow rate increases to a certain value, the distance of vapor cloud dispersion along the wind does not increase, or even decrease, but the width of the vapor cloud increases.
When the transfer flow rates are 500 m3/h, 1000 m3/h, 1500 m3/h, and 2000 m3/h, and the concentration reaches 25,000 ppm, the dispersion distance and heights are as shown in Figure 3. It can be seen from Figure 3 that with an increase in the transfer flow rate, the maximum downwind cloud height first increases, then decreases, and then increases again, and the change amplitude is small when the rate is large. Table 5 shows the heights of vapor cloud with different transfer flow rates.
When the transfer flow rate is 500 m3/h, the vapor cloud at night rises to a higher level for the second time, and the maximum height of the second rise is 8.35 m. With the increase in the transfer flow rate, the dispersion height of the vapor cloud does not change significantly.

3.2. Release Height

When the release height is 5 m, 10 m, 15 m, and 20 m, respectively, the dispersion distance and width of the vapor cloud with a concentration of 25,000 ppm are as shown in Figure 4. It can be seen from Figure 4 that, with an increase in release height, the dispersion distance of the vapor cloud in the downwind direction increases, and the dispersion distance in the upwind direction decreases; that is, the higher the release height, the more vulnerable the cloud movement direction is to the wind direction. Table 6 shows the dispersion distance and footprint for different release heights.
The change in release height significantly affects the initial dispersion characteristics of gas clouds. At night, the lower release height makes it easier for the gas to approach the water surface at the initial stage of release and is affected by the surface roughness and boundary layer turbulence. In this case, the vapor cloud disperses wider in the upwind direction because the turbulence near the water surface is more complex, resulting in the enhancement of the horizontal dispersion ability of the gas [53]. However, with an increase in release height, the distance between the gas release point and the water surface increases, and the vertical dispersion at the initial stage is more significant. A lower release height allows the vapor cloud to come into contact with the water surface faster. At a higher release height, the vapor cloud has more space to mix with the air, and after the temperature rises, it spreads further downwind, so the dispersion distance in the downwind direction gradually increases.
At the same release height, the downwind dispersion distance of release at night is greater than that of release during daytime. At night, when the release height is 5 m, the maximum downwind distance is 577.88 m, and the cloud footprint is 90,023 m2. When the height is 10 m, 15 m, and 20 m, respectively, the maximum downwind distance is 765.14 m, 1058.34 m and 1423.96 m, and the cloud footprint is 136,824 m2, 249,146 m2 and 468,881 m2. The cloud footprint at night shows a significant increase trend. However, with the increase in release height during the day, the maximum downwind distance of the vapor cloud increases slowly, and the width does not increase significantly. That is, the vapor cloud generated by the release at night spreads wider and farther in the downwind direction. The dispersion degree of the release vapor cloud at night is more vulnerable to the increase in release height because during the daytime, the higher release height provides a longer time for the gas to be driven by the wind speed, and the horizontal transmission capacity is stronger. This enhancement of horizontal dispersion is closely related to the weakening of the dilution effect of the gas cloud in the vertical direction. Due to the strong vertical turbulence, the vapor cloud with a higher height release forms a thinner concentration distribution, limiting the range of horizontal dispersion [54].
The release height also directly affects the width change of the vapor cloud. At the lower release height, the concentration of the gas cloud near the ground increases significantly, leading to a more significant dispersion of the cloud width in the upwind direction. On the contrary, at a higher release height, the dilution rate of gas in the higher air layer is accelerated, and the dispersion width in the upwind direction is smaller [55]. This can explain why at a height of 20 m, although the dispersion distance in the downwind direction is the largest, the cloud width in the upwind direction is still significantly reduced.
When the release heights are 5 m, 10 m, 15 m, and 20 m, and the concentration reaches 25,000 ppm, the dispersion distance and heights are as shown in Figure 5.
It can be seen from Figure 5 that with an increase in release height, the height of the vapor cloud near the release point increases. This is because the higher the release point is, the less chance the vapor cloud has to contact the water surface at the initial stage, and it is more likely to be affected by turbulence and wind speed, leading to upward dispersion. Table 7 shows the heights of vapor clouds with different release heights.
The dispersion height of the vapor cloud with release at night is higher than that in the daytime. The lower temperature at night and the lack of solar radiation cause the vapor cloud to form a more obvious temperature gradient in the vertical direction. The daytime vapor cloud starts to appear from about 50 m, mainly because the meteorological conditions in the daytime (high wind speed, unstable gas layer, strong solar radiation) promote rapid horizontal dispersion and turbulent mixing of vapor. These conditions make the vapor rapidly to dilute and disperse near the release point, preventing the formation of an obvious vapor cloud, but an obvious vapor cloud begins to form about 50 m downwind.
When the release height is 5 m, the maximum cloud height occurs during the second cloud rise, both during the day and at night. This is because at the initial stage of low altitude release, the vapor cloud is mainly affected by gravity and diffuses downward, and after forming a stable gas cloud layer within a certain range, it rises again after being far away from the release point due to horizontal transportation and turbulence. There is a certain difference in cloud height between day and night. The maximum cloud height during the day is 8.48 m, and the maximum cloud height at night is 10.54 m. This indicates that low wind speeds and an unstable atmosphere at night are more conducive to the vertical uplift of vapor clouds, and the turbulent effect is more significant at higher wind speeds during the day, resulting in a more uniform distribution of cloud height [56].
When the release height is 10 m, 15 m, and 20 m, the vapor cloud with release at night reaches the maximum height near the release point and the maximum height at the far release point in the daytime. This shows that at a higher release height, the vapor cloud is more vulnerable to wind speed. As the release height increases, the cloud height near the release point increases. This is because an increase in release height directly increases the potential height of the initial release of the gas cloud and enhances the ability of vertical dispersion. In addition, the higher the release height, the more the gas disperses in the higher air layers at the initial stage, reducing the degree of interaction with the water surface is reduced and resulting in a higher maximum cloud height [57,58]. With an increase in the release height, the height of the cloud secondary rise decreases, and the decrease is more significant at night. This is because clouds are already close to the stable or inversion layer in the atmosphere when released and are more likely to accumulate below the stable or inversion layer after the first lift, with limited space for the second lift.
The release height is a critical factor governing LNG dispersion behavior, where higher elevations promote downwind transport and vertical uplift, while lower heights enhance upwind spreading and ground-level accumulation. Nighttime conditions further amplify these effects due to reduced atmospheric mixing.

3.3. Release Direction

When the release direction is horizontal, 90° up, 90° down, and 45° down, respectively, the dispersion distance and width of the vapor cloud with a concentration of 25,000 ppm are as shown in Figure 6. It can be seen from Figure 6 that the release direction significantly affects the dispersion characteristics of the vapor cloud. When the release direction is vertical, the dispersion distance of the vapor cloud is short and the cloud footprint at night is small. The dispersion distance of the release vapor cloud in the horizontal direction is longer than that of the release vapor cloud in the 45° down direction, but the cloud width in the daytime is smaller. This phenomenon is closely related to the influence of the release direction on the initial momentum distribution of the gas cloud. Table 8 shows the dispersion distance and footprint for different release directions.
When the release direction is 90° up, the initial momentum of the gas is concentrated in the vertical direction, prompting the methane to be rapidly pushed to a higher air layer and mixed with the ambient air. In the daytime, methane is rapidly pushed to the atmosphere at a high altitude to form clouds in contact with air and water vapor. It then disperses along the downwind direction under the effect of wind, with the maximum downwind distance reaching 656.33 m. Under stable atmospheric conditions at night, turbulence in the vertical direction is weaker. The cloud formed above the release point of methane diffuses dispersion, and the distance along the downwind is 413.16 m, which is 37% shorter than the dispersion distance in the daytime.
When the release direction is 90° down, the cloud dispersion in the initial stage is closer to the ground. As the gas cloud is close to the water surface, the cloud dispersion distance along the wind is smaller than in other directions due to the influence of gravity and water surface roughness. The maximum downwind distance during the day is 521.72 m, and the maximum downwind distance at night is 375.79 m. However, when 90° down release occurs, the dispersion distance of the cloud in the upwind direction is larger than that in other directions.
When the release direction is horizontal, the initial momentum of the gas is consistent with the wind speed direction, which significantly increases the horizontal dispersion distance of the vapor cloud. Compared with the 45° down release vapor cloud, the downwind dispersion distance of the horizontal release vapor cloud is longer. The cloud footprint at night increased from 249,922 m2 to 458,254 m2, nearly doubling in size. The cloud footprint in the daytime is similar, but the cloud width in the daytime is small. Because in the horizontal release, the initial phase of the gas cloud is mainly driven by wind speed, its dispersion shape is narrower and more concentrated, and the proportion of dispersion in the vertical direction is lower.
When the release direction is 45° down, the dispersion behavior of the vapor cloud is between horizontal and 90° down release. Since the vapor cloud has part of the vertical momentum, its dispersion distance is slightly shorter than the dispersion distance of the horizontal release cloud, but the cloud width is slightly larger. This indicates that the angle of the release direction directly affects the initial dispersion path of the gas cloud and further affects its dispersion range and shape.
When the release direction is horizontal, 90° up, 90° down, and 45° down, the dispersion distance and height of the vapor cloud with a concentration of 25,000 ppm are as shown in Figure 7. It can be seen from Figure 7 that the release direction has no significant effect on the dispersion height of the cloud. Table 9 shows the heights of vapor clouds with different release directions.
At night, the vapor cloud reaches its maximum height near the release point. The maximum height of the release vapor cloud in the horizontal direction is 15.3 m. The maximum height of the release vapor cloud in the 90° up direction is 14.16 m. The maximum height of the release vapor cloud in the 90° down direction is 13.34 m. The maximum height of the release vapor cloud in the 45° down direction is 13.51 m. That is, the release of the transfer hose in different directions has little impact on the dispersion height of the vapor cloud under different meteorological conditions. However, the change in horizontal dispersion distance is consistent with the previous discussion. When the initial momentum of gas is consistent with the direction of wind speed, the horizontal dispersion distance of the vapor cloud increases significantly. It should also be noted that wind direction, wind speed, and atmospheric temperature are meteorological parameters that can significantly influence vapor cloud behavior. Wind direction determines the primary axis of cloud propagation. Higher wind speeds (e.g., daytime conditions) enhance horizontal transport but reduce cloud persistence due to increased turbulent mixing. Conversely, lower nighttime wind speeds allow for more concentrated and prolonged vapor accumulation. Atmospheric temperature gradients between cryogenic LNG (−162 °C) and ambient air further modulate dispersion through buoyancy effects, with warmer daytime temperatures promoting vertical uplift. These interactions highlight the importance of integrating real-time meteorological data into operational risk assessments.

4. Conclusions

This study simulates a series of scenarios of release during LNG filling operation and studies the influence of different key operational parameters on dispersion of an LNG vapor cloud. Simulating both daytime and nighttime conditions allows for the identification of various dispersion scenarios, with nighttime typically representing the higher-risk condition due to more stable atmospheric layers and limited vertical mixing. Using the unique weather and meteorological parameters in Shenzhen, Guangdong, China, the vapor cloud footprints of LNG release in the daytime and at night are compared.
The key findings are as follows:
Transfer flow rate:
  • Lower transfer flow rates reduce LNG leakage and rainout, improving safety.
  • Flow rates that are too low, however, increase bunkering time and reduce efficiency.
  • At night, increasing the flow rate beyond 1500 m3/h results in little change in the cloud footprint.
  • During the night, a flow rate of approximately 1500 m3/h should only be used when necessary, and strict monitoring is required.
  • Preferably, adopt a flow rate of 1000–1500 m3/h during the day to optimize risk control.
  • If conditions allow, using multiple low-flow hoses (e.g., three at 500 m3/h) is safer than using a single high-flow hose.
Release height:
  • Increasing the release height from 5 to 20 m significantly enlarges the cloud footprint, particularly by 1.5 times during the day and 5.2 times at night.
  • The inlet and outlet manifolds should be kept as close to 10 m as possible.
  • At a 20 m release height, the nighttime footprint is 3.5 times larger than the daytime footprint.
  • If the manifold height cannot be adjusted, bunkering should be done during the daytime, when upwind vapor dispersion is less likely.
  • Where feasible, ships should take in ballast water during bunkering to lower the intake manifold height.
  • LNG bunkering vessels should be equipped with low-level outlet manifolds to limit vapor dispersion.
Release direction:
  • Guiding LNG release vertically (90° downward) can reduce vapor cloud dispersion. If feasible, vertical release should be used, especially at night. This is to limit the dispersion area of the vapor cloud.
  • If the release direction cannot be controlled, filling should be done during the daytime.
  • When bunkering, the direction of release should not be the same as the wind direction.
  • The bunkering direction should align with the wind to avoid lateral dispersion. Alternatively, airflow can be managed using ventilation or protective systems.
  • When adding facilities, avoid forming confined spaces, as serious vapor cloud explosion may occur when LNG releases.
  • At night, use ventilation to guide the release downward; a 45° downward release reduces the footprint by 45% compared to a horizontal release.
This study reveals the influence mechanism of three operational parameters on LNG dispersion on seawater surface and quantifies the effect of limiting dispersion that can be achieved by parameter optimization. By identifying how each parameter influences the size and shape of the vapor cloud under varying conditions, this study provides a quantitative basis for assessing dispersion-related risk and prioritizing higher-risk scenarios for additional controls. The results show that the reasonable control of flow rate, height, and direction can effectively regulate the distance and range of LNG dispersion. These insights provide an important basis for safety design and operational procedure formulation of LNG fuel filling operations and offer practical guidance to enhance safety and environmental performance.
Additionally, the recommended safety measures have economic and operational implications. For instance, limiting the flow rate can increase bunkering time, which potentially means reduced daily throughput and higher operating costs and delays. However, this can be mitigated by using multiple low-flow hoses to maintain overall throughput. Adjusting bunkering direction or installing ventilation systems may involve additional upfront setup costs and time; however, these measures can prevent costly incidents, thereby reducing insurance risks and enhancing compliance. Therefore, these measures are practical and feasible from an operator’s perspective and can contribute to better risk management and help ensure operational continuity.
For future study directions, it is recommended to explore the following aspects: Firstly, conduct on-site tests on the actual release of LNG to verify the simulation results and enhance the credibility of the study conclusions. Secondly, expand the simulation scenarios and study the release patterns of LNG under different environmental conditions and storage states. Once again, breaking through the assumption of static LNG pools in the existing study, delve into the dispersion characteristics of underwater LNG and its movement mechanism with water flow. In addition, as this study focuses on vapor cloud dispersion resulting from LNG hose releases during STS bunkering operations, heat radiation effects were not considered. Future work should extend the simulation to include thermal radiation impacts for a more comprehensive risk assessment. Nighttime dispersion is likely further influenced by atmospheric stability and reduced thermal turbulence, which may suppress vertical mixing and prolong gas cloud persistence. Future work should incorporate detailed meteorological modeling (e.g., wind shear effects) to quantify these interactions. Beyond the proposed operational control measures, future work should also explore engineering solutions that aim to prevent or contain LNG leaks at the source, e.g., concentric double-tube transfer hoses, which can serve to contain leakages from inner hose. Advanced materials should be explored to minimize failure risks. Together with operational strategies, a more robust and comprehensive approach to LNG bunkering safety can be ensured. Finally, the influence of obstacles such as ship superstructure on LNG near-field dispersion can be systematically studied by using computational fluid dynamics.

Author Contributions

Conceptualization, Y.G. and Y.Z.; methodology, Y.G. and Y.Z.; validation, Y.G. and Y.Z.; formal analysis, Y.G., Y.Z., and H.S.L.; investigation, Y.G. and Y.Z.; resources, Y.G.; data curation, Y.G. and Y.Z.; writing—original draft preparation, Y.G., Y.Z., and H.S.L.; writing—review and editing, Y.G., Y.Z., and H.S.L.; visualization, Y.G.; supervision, Y.G.; project administration, Y.G. and Y.Z.; funding acquisition, Y.G. and Y.Z. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Natural Science Foundation of Guangdong Province, China, grant number 2023A1515010950; the Philosophy and Social Science Foundation of Guangdong Province, China, grant number GD22XGL03; and the Fundamental Research Funds for the Central Universities, China, grant number CXTD202407.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

Yanmin Zeng was employed by Shenzhen Port Group. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Carbon Pricing Dashboard. Available online: https://carbonpricingdashboard.worldbank.org/ (accessed on 12 November 2024).
  2. Boyland, J.; Talalasova, E. Anual Progress Report on Green Shipping Corridors; The Global Maritime Forum: Copenhagen, Denmark, 2024. [Google Scholar]
  3. General Administration of Customs of the People’s Republic of China. Available online: http://www.customs.gov.cn/ (accessed on 12 December 2024).
  4. Shenzhen Port Becomes the Fourth Global and First Hub Port in South China with LNG Refueling Service Capability. Available online: https://td.gd.gov.cn/gkmlpt/content/4/4051/mmpost_4051537.html#1479 (accessed on 12 December 2024). (In Chinese)
  5. Raj, P.K. LNG fires: A review of experimental results, models and hazard prediction challenges. J. Hazard. Mater. 2007, 140, 444–464. [Google Scholar] [CrossRef]
  6. Investigation Report on the “11.2” Major Fire Accident of Beihai Liquefied Natural Gas Co., Ltd. Available online: http://yjglt.gxzf.gov.cn/gwgg/t8366212.shtml (accessed on 20 November 2024). (In Chinese)
  7. Guidelines for LNG Fuel Bunkering Operation; China Classification Society: Beijing, China, 2021.
  8. Brown, T.C.; Cederwall, R.T.; Chan, S.T.; Ermak, D.L.; Koopman, R.P.; Lamson, K.C.; McClure, J.W.; Morris, L.K. Falcon Series Data Report: 1987 LNG Vapor Barrier Verification Field Trials; University of North Texas Libraries: Denton, TX, USA, 1990; 661p. [Google Scholar]
  9. Goldwire, H.C.; Rodean, H.C. 7 experiments with liquefied natural gas, 2 with liquid methane, and one with liquid nitrogen. In Coyote Series Data Report LLNL/NWC 1981 LNG Spill Tests Dispersion, Vapor Burn, and Rapid-Phase-Transition; Lawrence Livermore National Lab.: Livermore, CA, USA, 1983; Volume 1. [Google Scholar]
  10. Luketa-Hanlin, A. A review of large-scale LNG spills: Experiments and modeling. J. Hazard. Mater. 2006, 132, 119–140. [Google Scholar] [CrossRef] [PubMed]
  11. Koopman, R.P.; Ermak, D.L. Lessons learned from LNG safety research. J. Hazard. Mater. 2007, 140, 412–428. [Google Scholar] [CrossRef]
  12. Koopman, R.P.; Cederwall, R.T.; Ermak, D.L.; Goldwire, H.C.; Hogan, W.J.; Mcclure, J.W.; Mcrae, T.G.; Morgan, D.L.; Rodean, H.C.; Shinn, J.H. Analysis of Burro series 40-m3 LNG spill experiments. J. Hazard. Mater. 1982, 6, 43–83. [Google Scholar] [CrossRef]
  13. Cleaver, P.; Johnson, M.; Ho, B. A summary of some experimental data on LNG safety. J. Hazard. Mater. 2007, 140, 429–438. [Google Scholar] [CrossRef] [PubMed]
  14. Luketa-Hanlin, A.; Koopman, R.P.; Ermak, D.L. On the application of computational fluid dynamics codes for liquefied natural gas dispersion. J. Hazard. Mater. 2007, 140, 504–517. [Google Scholar] [CrossRef]
  15. Sun, B.; Utikar, R.P.; Pareek, V.K.; Guo, K. Computational fluid dynamics analysis of liquefied natural gas dispersion for risk assessment strategies. J. Loss Prev. Process Ind. 2013, 26, 117–128. [Google Scholar] [CrossRef]
  16. Zhang, Z.; Zhu, W.; Pan, X.; Tan, B.; Ye, Z.; Zhang, L.; Zhou, X.; Jiang, J.J.P.S.; Protection, E. Field experimental study of high expansion foam coverage after LNG leakage. Process Saf. Environ. Prot. 2024, 183, 315–326. [Google Scholar] [CrossRef]
  17. Hu, S.; Fang, Q.; Xia, H.; Xi, Y. Formal safety assessment based on relative risks model in ship navigation. Reliab. Eng. Syst. Saf. 2007, 92, 369–377. [Google Scholar] [CrossRef]
  18. Hissong, D.W. Keys to modeling LNG spills on water. J. Hazard. Mater. 2007, 140, 465–477. [Google Scholar] [CrossRef]
  19. Lehr, W.; Simecek-Beatty, D. Comparison of hypothetical LNG and fuel oil fires on water. J. Hazard. Mater. 2004, 107, 3–9. [Google Scholar] [CrossRef] [PubMed]
  20. Sun, W.; Yang, S.; Yang, Y.; Fang, Q.; Zhu, J.; Dai, J.; Li, H. Investigation on the concentration prediction model and personnel hazard range of LNG leakage from tankers in the tunnel. Process Saf. Environ. Prot. 2023, 172, 700–715. [Google Scholar] [CrossRef]
  21. Zhang, Q.; Wu, J.; Bai, Y.; Zhang, C.; Wang, J.; Qin, T. Agent-based risk modeling of domino effects in urban LNG stations. Journal of Loss Prevention in the Process Industries 2024, 89, 105300. [Google Scholar] [CrossRef]
  22. Nubli, H.; Sohn, J.-M.; Jung, D. Consequence Analysis of Accidental LNG Release on the Collided Structure of 500 cbm LNG Bunkering Ship. J. Mar. Sci. Eng. 2022, 10, 1378. [Google Scholar] [CrossRef]
  23. Luan, X.; Zhang, M.; Zhao, S.; Zhang, B. Numerical study on the effects of bund on liquid pool spreading and vapor dispersion after a catastrophic LNG tank failure. Process Saf. Environ. Prot. 2023, 176, 74–86. [Google Scholar] [CrossRef]
  24. Xie, Q.; Lu, Q.; Yuan, Y.; Zhang, J.; Zhou, F. Numerical study on the horizontal stretching effect of ground on high-pressure vapor jets of LNG tank leakage. J. Loss Prev. Process Ind. 2021, 72, 104526. [Google Scholar] [CrossRef]
  25. Chan, S.T. Numerical simulations of LNG vapor dispersion from a fenced storage area. J. Hazard. Mater. 1992, 30, 195–224. [Google Scholar] [CrossRef]
  26. Hankin, R.K.S. Heavy gas dispersion: Integral models and shallow layer models. J. Hazard. Mater. 2003, 103, 1–10. [Google Scholar] [CrossRef]
  27. Animah, I.; Shafiee, M. Application of risk analysis in the liquefied natural gas (LNG) sector: An overview. J. Loss Prev. Process Ind. 2020, 63, 103980. [Google Scholar] [CrossRef]
  28. Johnson, D.W.; Cornwell, J.B. Modeling the release, spreading, and burning of LNG, LPG, and gasoline on water. J. Hazard. Mater. 2007, 140, 535–540. [Google Scholar] [CrossRef]
  29. Fay, J.A. Model of spills and fires from LNG and oil tankers. J. Hazard. Mater. 2003, 96, 171–188. [Google Scholar] [CrossRef]
  30. Zhang, K.; Zhou, X.; Cong, B. The effect of relative humidity on vapor dispersion of liquefied natural gas: A CFD simulation using three phase change models. J. Wind Eng. Ind. Aerodyn. 2022, 230, 105181. [Google Scholar] [CrossRef]
  31. Węsierski, T.; Piec, R.; Majder-Łopatka, M.; Król, B.; Gawroński, W.; Kwiatkowski, M. Hazards Generated by an LNG Road Tanker Leak: Field Investigation of Vapour Propagation under Class B Conditions of Atmospheric Stability. Energies 2021, 14, 8483. [Google Scholar] [CrossRef]
  32. Bozzolasco, M. LNG Terminal Risk Assessment Consequence Analysis of Accident Scenarios. In Proceedings of the Offshore Mediterranean Conference and Exhibition, Ravenna, Italy, 16–18 March 2005. [Google Scholar]
  33. Wang, K.; Liu, Z.; Qian, X.; Huang, P. Long-term consequence and vulnerability assessment of thermal radiation hazard from LNG explosive fireball in open space based on full-scale experiment and PHAST. J. Loss Prev. Process Ind. 2017, 46, 13–22. [Google Scholar] [CrossRef]
  34. Dan, S.; Lee, C.J.; Park, J.; Shin, D.; Yoon, E.S. Quantitative risk analysis of fire and explosion on the top-side LNG-liquefaction process of LNG-FPSO. Process Saf. Environ. Prot. 2014, 92, 430–441. [Google Scholar] [CrossRef]
  35. Zhu, M.; Huang, L.; Huang, Z.; Shi, F.; Xie, C.J.O.; Management, C. Hazard analysis by leakage and diffusion in Liquefied Natural Gas ships during emergency transfer operations on coastal waters. Ocean Coast. Manag. 2022, 220, 106100. [Google Scholar] [CrossRef]
  36. Xie, C.; Huang, L.; Deng, J.; Wang, R.; Hao, G. Hazard assessment and hazard mitigation of fuel leak inside a ship elevator for LNG-fueled vessel. Ocean Eng. 2022, 259, 111943. [Google Scholar] [CrossRef]
  37. Zhang, Z.; Krishnan, P.; Jiao, Z.; Mannan, M.S.; Wang, Q. Developing a CFD heat transfer model for applying high expansion foam in an LNG spill. J. Loss Prev. Process Ind. 2021, 71, 104456. [Google Scholar] [CrossRef]
  38. Lin, Z.; Wang, S.; Fu, S.; Huo, J. Numerical Study on Effects of the Cofferdam Area in LNG Storage Tank on the Leakage and Diffusion Characteristics of Natural Gas. Chin. J. Chem. Eng. 2021, 29, 228–241. [Google Scholar] [CrossRef]
  39. Duong, P.A.; Ryu, B.R.; Jung, J.; Kang, H. Comparative analysis on vapor cloud dispersion between LNG/liquid NH_3 leakage on the ship to ship bunkering for establishing safety zones. J. Loss Prev. Process Ind. 2023, 85, 105167. [Google Scholar] [CrossRef]
  40. Cao, Y.; Jia, Q.-J.; Wang, S.-M.; Jiang, Y.; Bai, Y. Safety design analysis of a vent mast on a LNG powered ship during a low-temperature combustible gas leakage accident. J. Ocean Eng. Sci. 2022, 7, 75–83. [Google Scholar] [CrossRef]
  41. Àgueda, A.; Subirana, J.; Pastor, E.; Schleder, A.M.; Eulàlia, P. Revisiting the dispersion safety factor (DSF) for vapor clouds of liquefied flammable gases (LNG and propane). Saf. Sci. 2020, 128, 104748. [Google Scholar] [CrossRef]
  42. Gerbec, M.; Vidmar, P.; Pio, G.; Salzano, E. A comparison of dispersion models for the LNG dispersion at port of Koper, Slovenia. Saf. Sci. 2021, 144, 105467. [Google Scholar] [CrossRef]
  43. DNV. DNV’s Spadeadam Research and Testing Facility. Available online: https://www.dnv.com/energy/services/laboratories-test-facilities/spadeadam/ (accessed on 20 November 2024).
  44. Duong, P.A.; Ryu, B.R.; Jung, J.; Kang, H. A Comprehensive Review of the Establishment of Safety Zones and Quantitative Risk Analysis during Ship-to-Ship LNG Bunkering. Energies (19961073) 2024, 17, 512. [Google Scholar] [CrossRef]
  45. Ryu, B.R.; Duong, P.A.; Kim, J.-B.; Choi, S.-Y.; Shin, J.W.; Jung, J.; Kang, H. The Effect of Ventilation on the Hazards of Hydrogen Release in Enclosed Areas of Hydrogen-Fueled Ship. Journal of Marine Science and Engineering 2023, 11, 1639. [Google Scholar] [CrossRef]
  46. Fan, H.; Enshaei, H.; Jayasinghe, S.G. Human Error Probability Assessment for LNG Bunkering Based on Fuzzy Bayesian Network-CREAM Model. J. Mar. Sci. Eng. 2022, 10, 333. [Google Scholar] [CrossRef]
  47. Feng, W.; Wang, Z.; Dai, X.; Dong, S.; Qiao, W.; Ma, X. Ship-To-Ship Liquefied Natural Gas Bunkering Risk Assessment by Integrating Fuzzy Failure Mode and Effect Analysis and the Technique for Order Preference by Similarity to an Ideal Solution. J. Mar. Sci. Eng. 2025, 13, 710. [Google Scholar] [CrossRef]
  48. Guidelines for Quantitative Risk Assessment: Purple Book; Committee for the Prevention of the Disasters: The Netherland, 2004.
  49. Wolting, A.G.; Vijgen, L.J. Development of a QRA method to calculate the risks generated by Liquefied Natural Gas (LNG) filling stations for road trucks. In Proceedings of the AIChE Spring Meeting and 9th Global Congress on Process Safety, San Antonio, TX, USA, 28 April–2 May 2013. [Google Scholar]
  50. Liang, T.; Chen, G. Application Research of SAFETI on the Fatal Accident Consequence Assessment in LPG Storage Tanks. Oil Gas Storage Transp. 2006, 25, 53–58. [Google Scholar]
  51. Meteorological Bureau of Shenzhen Municipality. Available online: https://weather.sz.gov.cn/ (accessed on 10 October 2024).
  52. OCIMF. Guide to Manufacturing and Purchasing Hoses for Offshore Moorings, 5th ed.; Oil Companies International Marine Forum: London, UK, 2009. [Google Scholar]
  53. Hanna, S.R.; Chang, J.C.; Strimaitis, D.G. Hazardous gas model evaluation with field observations. Atmos. Environment. Part A Gen. Top. 1993, 27, 2265–2285. [Google Scholar] [CrossRef]
  54. Brzezinska, D.; Markowski, A.S. Experimental evaluation of LPG release and dispersion in the enclosed car parks. Chem. Eng. Trans. 2016, 48, 253–258. [Google Scholar]
  55. Markiewicz, M.T. Mathematical Modeling of the Heavy Gas Dispersion. In Models and Techniques for Health and Environmental Hazardassessment and Management, MANHAZ Monograph; Institute of Atomic Energy: Otwock-Świerk, Poland, 2006. [Google Scholar]
  56. Iman, R.L.; Helton, J.C.; Johnson, J.D. A Methodology for Grouping Source Terms for Consequence Calculations in Probabilistic Risk Assessments. Risk Anal. 2010, 10, 507–520. [Google Scholar] [CrossRef]
  57. Cornwell, J.B.; Johnson, D.W. Modeling LNG spills on water. In Proceedings of the 4th Topical Conference on Natural Gas Utilization at the 2004 AIChE National Spring Meeting, New Orleans, LA, USA, 25–29 April 2024. [Google Scholar]
  58. Woodward, J.L.; Pitblado, R. LNG Risk Based Safety: Modeling and Consequence Analysis; Wiley-AIChE: Hoboken, NJ, USA, 2012. [Google Scholar]
Jmse 13 01494 g001
Figure 1. Ship-to-ship (STS) bunkering mode [7].
Figure 1. Ship-to-ship (STS) bunkering mode [7].
Jmse 13 01494 g001
Jmse 13 01494 g002
Figure 2. Horizontal dispersion and footprint (25,000 ppm) of vapor clouds with different transfer flow rates. Source: Compiled by the authors.
Figure 2. Horizontal dispersion and footprint (25,000 ppm) of vapor clouds with different transfer flow rates. Source: Compiled by the authors.
Jmse 13 01494 g002
Jmse 13 01494 g003
Figure 3. Vertical dispersion for different transfer flow rates. Source: Compiled by the authors.
Figure 3. Vertical dispersion for different transfer flow rates. Source: Compiled by the authors.
Jmse 13 01494 g003
Jmse 13 01494 g004
Figure 4. Horizontal dispersion and footprint (25,000 ppm) for different release heights. Source: Compiled by the authors.
Figure 4. Horizontal dispersion and footprint (25,000 ppm) for different release heights. Source: Compiled by the authors.
Jmse 13 01494 g004
Jmse 13 01494 g005
Figure 5. Vertical dispersion for different release heights. Source: Compiled by the authors.
Figure 5. Vertical dispersion for different release heights. Source: Compiled by the authors.
Jmse 13 01494 g005
Jmse 13 01494 g006
Figure 6. Horizontal dispersion and footprint (25,000 ppm) of vapor clouds with different release directions. Source: Compiled by the authors.
Figure 6. Horizontal dispersion and footprint (25,000 ppm) of vapor clouds with different release directions. Source: Compiled by the authors.
Jmse 13 01494 g006
Jmse 13 01494 g007
Figure 7. Vertical dispersion for different release directions. Source: Compiled by the authors.
Figure 7. Vertical dispersion for different release directions. Source: Compiled by the authors.
Jmse 13 01494 g007
Table 1. Comparison of three hazard analysis tools.
Table 1. Comparison of three hazard analysis tools.
ToolsPHASTFluentFLACS
Computational SpeedVery fast
Suitable for rapid assessment		Slower
Suitable for detailed simulation		Moderate
Suitable for explosion and fire simulation
User-FriendlinessUser-friendly
Intuitive interface		More complex
Requires CFD expertise		Moderate
Requires some training
Scope of ApplicationLarge-scale scenarios
Suitable for plant-level assessment		Localized detailed simulationComplex geometric environments for explosion simulation
Resource RequirementsLow
Minimal hardware requirements		High
Requires high-performance computing resources		Moderate
Requires certain computing resources
ComprehensivenessComprehensive
Includes leakage, dispersion, fire, and explosion		Focuses on fluid flow and heat transferFocuses on explosion and fire simulation
Cost-EffectivenessHigh
Suitable for budget-limited projects		Fairly high, due to software and hardware costsModerate
Table 2. Typical meteorological data [51].
Table 2. Typical meteorological data [51].
Weather ParameterDayNight
Pasquill StabilityCE
Wind speed (m/s)32
Ambient Temperature (°C)2319
Relative Humidity (%)7685
Solar Radiation Flux (kW/m2)10
Sea Surface Temperature (°C)2519
Table 3. Operational conditions for model inputs.
Table 3. Operational conditions for model inputs.
CategoryDescriptionOrifice Size (mm)Storage ConditionFlow Rate
(m3/h)		Release Height
(m)		Release Direction
ADifferent flow rates20−162 °C, 101 kPa500, 1000, 1500, 200018.3horizontal
BDifferent release heights10005, 10, 15, 20horizontal
CDifferent release directions100018.3horizontal, 90° up,
90° down, 45° down
Table 4. Dispersion distance and footprint for different transfer flow rates.
Table 4. Dispersion distance and footprint for different transfer flow rates.
Transfer Flow RateDayNight
Maximum Downwind Distance (m)Maximum Upwind Distance (m)Cloud Footprint (m2)Maximum Downwind Distance (m)Maximum Upwind Distance (m)Cloud Footprint (m2)
500 m3/h429.05-33,870530.36−26.0254,719
1000 m3/h787.22128,9031610.13−25.38458,254
1500 m3/h1003.74237,0361493.49−22.84654,081
2000 m3/h1635.66485,6561416.16−17.31666,580
Source: Compiled by the authors.
Table 5. Heights of vapor cloud with different transfer flow rates.
Table 5. Heights of vapor cloud with different transfer flow rates.
Transfer Flow RateDayNight
Location of Maximum Height (m)Maximum Downwind Cloud Height (m)Location of Maximum Height (m)Maximum Downwind Cloud Height (m)
500 m3/h271.115.683.6412.75
1000 m3/h533.005.301.5915.30
1500 m3/h249.125.576.6812.65
2000 m3/h274.826.3413.5913.69
Source: Compiled by the authors.
Table 6. Dispersion distance and footprint for different release heights.
Table 6. Dispersion distance and footprint for different release heights.
Release HeightDayNight
Maximum Downwind Distance (m)Maximum Upwind Distance (m)Cloud Footprint (m2)Maximum Downwind Distance (m)Maximum Upwind Distance (m)Cloud Footprint (m2)
5 m430.95-54,475577.88−41.8790,023
10 m568.0678,130765.14−35.78136,824
15 m684.75113,6101058.34−11.41249,146
20 m799.15135,5991423.96−22.42468,881
Source: Compiled by the authors.
Table 7. Heights of vapor clouds with different release heights.
Table 7. Heights of vapor clouds with different release heights.
Release HeightDayNight
Location of Maximum Height (m)Maximum Downwind Cloud Height (m)Location of Maximum Height (m)Maximum Downwind Cloud Height (m)
5 m288.268.48254.1510.54
10 m366.457.120.649.40
15 m464.705.511.6812.59
20 m521.604.973.4314.11
Source: Compiled by the authors.
Table 8. Dispersion distance and footprint for different release directions.
Table 8. Dispersion distance and footprint for different release directions.
Release DirectionDayNight
Maximum Downwind Distance (m)Maximum Upwind Distance (m)Cloud Footprint (m2)Maximum Downwind Distance (m)Maximum Upwind Distance (m)Cloud Footprint (m2)
Horizontal787.22-128,9031610.13−25.38458,254
90° up669.47123,303413.16−16.52110,972
90° down521.7292,227375.79−46.36100,624
45° down675.76124,531967.92−35.06249,922
Source: Compiled by the authors.
Table 9. Heights of vapor clouds with different release directions.
Table 9. Heights of vapor clouds with different release directions.
Release DirectionDayNight
Location of Maximum Height (m)Maximum Downwind Cloud Height (m)Location of Maximum Height (m)Maximum Downwind Cloud Height (m)
Horizontal533.61 5.30 1.59 15.30
90° up32.53 6.30 0.08 14.16
90° down403.96 6.33 −0.59 13.34
45° down458.85 5.29 2.70 13.51
Source: Compiled by the authors.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Gu, Y.; Zeng, Y.; Loh, H.S. Quantitative Risk Assessment of Liquefied Natural Gas Bunkering Hoses in Maritime Operations: A Case of Shenzhen Port. J. Mar. Sci. Eng. 2025, 13, 1494. https://doi.org/10.3390/jmse13081494

AMA Style

Gu Y, Zeng Y, Loh HS. Quantitative Risk Assessment of Liquefied Natural Gas Bunkering Hoses in Maritime Operations: A Case of Shenzhen Port. Journal of Marine Science and Engineering. 2025; 13(8):1494. https://doi.org/10.3390/jmse13081494

Chicago/Turabian Style

Gu, Yimiao, Yanmin Zeng, and Hui Shan Loh. 2025. "Quantitative Risk Assessment of Liquefied Natural Gas Bunkering Hoses in Maritime Operations: A Case of Shenzhen Port" Journal of Marine Science and Engineering 13, no. 8: 1494. https://doi.org/10.3390/jmse13081494

APA Style

Gu, Y., Zeng, Y., & Loh, H. S. (2025). Quantitative Risk Assessment of Liquefied Natural Gas Bunkering Hoses in Maritime Operations: A Case of Shenzhen Port. Journal of Marine Science and Engineering, 13(8), 1494. https://doi.org/10.3390/jmse13081494

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop